Skip to the stories
AI & ML

Microsoft Outlook Outage Traced to Network Capacity Upgrade and Server Health Checks

  • Richard Brown
  • 962 reads

Microsoft linked the widespread Microsoft 365 disruption to a network update and a capacity upgrade. A full technical report is still pending.

Microsoft Outlook Outage Traced to Network Capacity Upgrade and Server Health Checks

IT worldwide outrage on 19th July 2024, blue screen on laptop. Photo is made in Spain, Marbella - stock photo

Microsoft has tied a widespread outage affecting Outlook and other Microsoft 365 services to a recent network change that exposed an earlier problem during a capacity upgrade. The company said it had addressed the underlying issue and was restoring service, but its explanation leaves open how the change affected server health checks and why the disruption spread across multiple regions.

That distinction matters for customers and IT teams. Microsoft’s initial account focused on servers becoming unavailable and struggling under heavier traffic. Its later statement identified a network change as the trigger. The company has promised a full post-incident report, which should clarify how those pieces fit together.

How the failure spread across services

The incident began around 12:30 p.m. ET on Thursday, Nov. 12. Microsoft first reported problems accessing Exchange Online mailboxes in the Americas. Exchange Online stores and delivers email for many Outlook users, so trouble in that cloud service can affect the app even when the app itself has no identified fault.

The reported impact soon extended beyond email. Users experienced problems with Outlook, Teams, SharePoint Online, Copilot, Defender XDR and other Microsoft 365 services. Microsoft later warned that connectivity problems could affect users in any region. Azure appeared to remain operational, according to an update reported during the outage.

By 1 p.m. ET, Downdetector had received nearly 130,000 user error reports. People described difficulty loading inboxes and sending or receiving messages, while Copilot users also reported trouble. That report count reflects user submissions to the service; it does not establish how many customers were affected in total.

Exchange Online is part of the cloud infrastructure behind several user-facing Microsoft products. When that shared layer has trouble, symptoms can appear in multiple apps at once. The spread of reports across services therefore pointed to a problem deeper in the service stack, though early updates did not identify its precise cause.

Servers, health checks and a network change

Microsoft’s first updates said some servers had become “unexpectedly unavailable.” The company also described heavier-than-usual load across parts of its infrastructure in multiple regions, which it believed was adding to connection problems.

Another early clue involved a component responsible for processing server health. Microsoft said servers were going offline after failing automated health checks. Those checks help determine whether a server is available to handle work. If they fail, servers may be taken out of service; if too many become unavailable while demand is elevated, remaining capacity can come under added pressure.

Microsoft initially said it had not found a recent change that lined up with the outage. It tried rolling back recent server changes and balancing traffic across its network. When those steps did not produce the expected results, the company continued investigating why the checks had failed and adjusted traffic going to affected servers.

On Friday morning, Microsoft gave a more specific explanation: the incident was introduced by an issue during an earlier network capacity upgrade, which surfaced after a recent network change. The company said it had addressed the underlying issue and was working to restore service to remaining customers and regions.

This account identifies a trigger, but not the full failure chain. It does not explain what went wrong in the capacity upgrade, what the later network change altered, or why the health-check component responded as it did. Those details will matter for understanding whether the problem came from the upgrade itself, how it was activated, or the way the system detected and handled failures.

Recovery took time

Microsoft’s attempts to restore servers did not immediately work as planned. The company reported that changes intended to improve serviceability were not behaving as expected. Some servers that returned to service could become unresponsive again when traffic reached them, so Microsoft limited the types of traffic sent to those servers in some locations.

By Thursday evening, Microsoft said some servers it had brought back online remained stable and warned that recovery would be gradual. At 7:09 a.m. ET on Friday, it reported improving service availability and email delivery across a growing portion of affected infrastructure. Users could still encounter intermittent problems while recovery actions continued.

By Friday afternoon, Microsoft described broad recovery and stability across its services. Charles Lamanna, Microsoft’s executive vice president for Copilot, Agents and Platform, said the company’s teams were reviewing what happened and how to prevent similar issues. Microsoft also apologized to customers and partners, acknowledging the disruption to people who rely on the services for daily work.

What the company has ruled out

As the disruption continued into a second day, Microsoft addressed speculation about a cyberattack or AI agents. The company described the event as a technical issue unrelated to cyber activity or AI agents. It also said it was not aware of malicious activity causing the incident.

Those statements narrow the explanation toward infrastructure and network operations. They do not replace the promised technical report, which should provide a clearer account of the failed health checks, the role of traffic load and the steps taken to restore capacity.

For organizations that depend on Microsoft 365, the practical concern is the combination of a shared service dependency and a recovery that unfolded over many hours. The next useful detail will be how Microsoft plans to catch this kind of failure earlier and keep server recovery stable under returning traffic. Until the full report arrives, the network change is the clearest identified trigger, while the exact mechanism remains unresolved.

Source: Richard Brown · mashable.com

From the same set